Devin, the autonomous AI developer from Cognition Labs that we covered back in March, is now finally available, and this week engineering teams all around the world, including ExoBrain’s, have been on-boarding the shape of things to come. So far, we’ve been impressed; Devin is a capable and thorough team member, and most of all, is very willing to learn. With a clever memory feature, plus configurable playbooks, we can see Devin becoming ever more familiar with our projects, and able to handle increasingly complex tasks. We’ve been integrating Devin into our workflow via Cursor’s IDE extension, which keeps us updated on its activity, and Slack, where it can manage multiple conversations at once. Despite its strengths, Devin still benefits from super clear guidance and well-scoped tasks to ensure it stays on track. Its API opens up even more potential, such as programmatically launching instances to execute security patches or automate specific workflows.
With models improving and costs reducing, these autonomous colleagues can only improve; this is a glimpse into the future of many technical jobs. Devin is also part of a bigger picture. There is a growing ecosystem of AI tools shaping a new way to build software. With tools like Devin, Gemini Deep Research, Pinokio, o1 Pro Mode, Bolt, and Cursor working in tandem, it’s possible to employ an AI-enabled assembly line where ideas flow rapidly from strategy to deployment:
- The newly released Gemini Deep Research (which uses a web browsing agent to source tens of documents and pages before writing a report in Google Docs), plus Perplexity’s AI search start the process with their ability to support market research and concept exploration.
- OpenAI’s o1 Pro Mode can significantly extend these early ideas with strategic insights and detailed architectural planning and design.
- Pinokio, meanwhile, is a useful piece of the puzzle, offering a quick way to find, install, and test the latest AI tools and open-source projects.
- Bolt.new accelerates prototyping in the browser, leveraging Claude’s coding capabilities and web containers, ideas can be realised in a fraction of the typical time to create a testable prototype.
- Cursor is emerging as the AI development powerhouse (with Windsurf hot on its heels). It facilitates accelerated coding with its integrated multi-model chat, slick in-line AI editing, and composer agents that can create and edit files autonomously. It also seamlessly pulls in external documentation, design documents, and coding standards to ensure the AI development is informed and aligned.
- Vercel’s v0 is the AI-powered tool changing how developers create user interfaces. It offers a chat-based approach to generate production-ready UI components using modern web technologies like Next.js and React.
And finally, Devin can handle the heavy or time-consuming work once your product is up and running, testing new features, making repetitive changes, analysing bugs, and many of the hard yards needed to keep software up to date and robust over the longer term.
Takeaways: Devin and a range of other AI tools are reshaping how software is built. When combined thoughtfully, they can speed up development from initial research through to deployment and maintenance. While software engineering is an early testing ground, the lessons learned here will help guide AI adoption across many other technical fields.